James Sinclair
James Sinclair

A flash in the pan or a systemic change?

James Sinclair, executive chairman at MarketFactory, sets out to define a flash event, ask how frequent they are, determine what factors cause them and how market participants may minimise their impact

Christmas decorations still decked the halls, children were not yet back at school and the first insults of the year had not yet been hurled in the fractured politics of our time – yet, the FX press had already yelled “flash crash!”   On January 3rd 2019, the yen experienced a flash event.  This led many traders to ask if this is the “new normal” and customers to ask how they can protect themselves from being on the wrong side of a flash event. Flash events have been with us for many years with the October 2016 GBP flash event attracting much attention due to its impact. 

Defining a Flash Event

Several definitions have been proposed:

1. MarketFactory:  Large, fast, price move (5 standard deviations or more) followed by illiquidity and/or divergent pricing across venue
2. Pramga Securities: a large price move (>13× normal volatility), strong reversion (>70%), and widening spread (>2× normal Spread)
3. BIS:  (relating specifically to the 2016 Sterling event): a large, fast, v-shaped price move and sudden widening of bid-offer spreads

How frequent are flash events? And do they matter anyway?

Most published metrics are biased towards not just the frequency but also the impact.  After all, if a flash event inconveniences nobody – like George Berkley’s tree falling in the woods - why should we care?  The answer is that we should care because, if there was a flash event someone must have traded and that someone had a 50% chance of receiving sub-optimal execution, whether they knew it or not. The absolute number of flash events is significant.

The Pragma paper reports that they found 69 events across calendar years 2015 and 2016 when studying  AUD, EUR, GBP, CAD, CHF, and JPY. This is of course vastly more than those that are generally known.

A summary of what we learned is below:
A summary of what we learned is below:
Time lapse of two hour crash view
Time lapse of two hour crash view

Another clue regarding the frequency of flash events may work by Kim Christensen, Roel Oomen and Roberto Renò in their September 2016 paper The Drift Burst Hypothesis.  Roel Oomen is of course well-known to us in FX for his many other papers and his role as a Managing Director at Deutsche Bank. In their paper the authors state that flash events -

“… are an expected and regular occurrence in financial markets that can arise through established mechanisms of liquidity provision”.  In effect, the authors expect that by its very nature, market  making leads to flash events. The net result is that flash events are more frequent than we may realise - and they do matter.

The Structure of a flash event.

The BIS summarized the Sterling flash event in three stages, and these stages are remarkably similar in almost all flash events we have analyzed:

Stage 1: (“Depreciation”) Shortly after midnight British Summer Time (BST), equivalent to Greenwich Mean Time (GMT) plus one hour, on 7 October 2013, trading volumes picked up sharply and sterling began to depreciate against other currencies.
Stage 2: (“Disorder”) At 00:07:15 BST, the CME triggered its velocity logic mechanism, which pauses trading for 10 seconds on the futures exchange, in response to the large moves in the preceding two seconds. At this point, bid-offer spreads in the spot market widened significantly. It is not possible to disentangle the impact of the trading pause on the CME, if any, on spot market liquidity from the impact of the preceding large price moves.
Stage 3:  (“Recovery”) Over time the market began to recover, although it is difficult to identify a clear shift to the recovery phase. By around 00:20:00 BST, prices in both the futures and spot market had settled around 2.2% lower against the dollar than their levels immediately prior to the event.

Only traders looking at these ECNs saw this trend

Only traders looking at these ECNs saw this trend

Comparing the size of flash events

It is true that all flash events are not equal. Comparing the Sterling flash event of July 6, 2016 with the yen event of Jan 3, 2019 we see a stark contrast not immediately apparent from the press reports:

Comparing the yen flash event to the sterling flash event

Comparing the size of flash events

What caused the events?

The 2016 sterling event remains somewhat opaque.   Some say they have traced it down to a fat finger error.  Others attribute the issue to options.  It is true that there are often (if not usually) large exotic option positions left overnight.   The 2019 yen event appears, at least to me to be much clearer.  

Yen Flash Event – Did the Tail Wag the Dog?

What caused the yen flash event?   Could it really have been a retail-led move following a negative revenue announcement by Apple? I agree it was surprising, but the evidence that the move was retail-led is convincing and I believe it was the case.

As others have observed, the clue is in which currencies moved and which did not. Currencies that moved sharply included the Mexican peso, South African rand and Turkish lira. There was no news from those countries.

These are all currencies that are especially associated with retail trading. Currencies that barely moved included the Swiss franc, a currency that usually appreciates in a stress event as institutional investors seek safe havens currencies.

Other circumstances contributing to this move included: 1) it took place during the Japanese holiday when there would have been less Japanese institutional participation, and 2) it took place during the one hour period when the CME is down each day (southeast Asian institutions that trade on CME may otherwise have provided some resistance to the move). Add the usual pre-existing exotic options and you have a flammable cocktail.

Some commentators have expressed disbelief that the move was retail-led because retail trading represents only about 4% of global FX trading. This ignores the disproportionate number of retail traders in Japan, who are leveraged at up to 25x.
A November 2018 paper by Yui Mukoyama, Naoya Kikuta, and Kazuaki Washimi of The Bank of Japan reports that 70% of retail investors are trend-followers. Trend followers would therefore amplify any move.

The paper goes on to say that even the contrarians are not strictly so and may become trend followers in times of volatility, which the Apple revenue announcement on January 2 certainly caused.

I agree that it is surprising that such a move could have been retail led. It is somewhat humbling for those of us who have worked in institutional foreign exchange for most of our lives. The tail really did wag the do.

Why are flash events becoming more common?

There are perhaps several reasons:

1. There are simply fewer counterparties. Japan.  A massive wave of bank mergers among the “city banks” in the the 1990s and 2000s have resulted in far fewer banks.  They have consolidated principally into Mitsubishi UFJ, Mizuho and Mitsui.   There appears to be more appetite now for mergers among the regional banks, beginning with 18th bank, which could further reduce the number of counterparties.

2. Fewer US and European banks appear to be active overnight,  They either outsource their overnight customer business using cover-and-deal mechanisms to a few market makers.  Others, with less customer business, don’t trade overnight, use stop losses and buy protection through the options market.

3. There is an increasing abundance of live market data.  It used to be that a dealer  had to trade to enable proce discovery and sensing the depth.  There is little need to trade for top of book price discovery and it is increasingly possible to infer the depth without trading.   This enables traders to engage in a passive rather than active trading.  

The net result is, at least, that flash events do indeed represent a “new normal”

So, how can traders avoid being on the “wrong side” of a flash event?   Ultimately, if you don’t have to trade at illiquid times, don’t enter the market at those times. Buy protection from sharp moves through the options market and through stop losses.  Neither course of action is good for the market overall, but it is good for each one individually.